Myelodysplastic syndromes (MDS) constitute a group of heterogeneous bone marrow stem cell disorders where no single treatment strategy has been of universal benefit. Several drugs have recently been approved by the FDA for MDS, but except for the subgroup of patients with a del(5q) karyotype, it is difficult to know which of the therapeutic options are likely to be of greatest benefit to the individual patient. All of the drugs have side effects that can be deleterious to the patient; knowing which drug is likely to improve the cytopenia(s) of the individual patient would thus be highly desirable. Gene expression profiling has been used to stratify patients into biologically similar subgroups. We hypothesized that patients with a similar natural history and response to a specific drug therapy may share an expression profile that distinguishes between responders and non- responders. Using this technology, we compared gene expression profiles of patients who responded to lenalidomide to those who did not, and identified a unique signature associated with response. Responders to lenalidomide under-expressed genes associated with erythroid differentiation, suggesting a defect in the ability to produce mature erythrocytes. We then validated this expression signature in an independent group of patients. The defective erythroid signature could be used to predict sensitivity to lenalidomide (quantified by a calculated z-score), providing a molecular validation of the clinical observation that this drug specifically benefits anemia in MDS patients. In the current grant, we propose to further investigate the ability to predict patient response using the erythroid expression signature. We will conduct a prospective trial of non-del(5q) transfusion dependant low/Int-1 MDS patient treated with single agent lenalidomide. Expression profiling by both microarray and luminex assay will be performed on the pre-treatment marrow and correlated with clinical response. If necessary, it will be possible to refine both the signature and the boundaries of the z-score with the data obtained. The additional patient numbers will enable the generation of a receiver operating characteristic curve that measures how well the signature separates responder from non-responder and gives a probability of response. The results of this study are likely to provide a clinical tool that could be used to pre-select patients likely to respond to the drug thus sparing the others from undesirable toxic exposure. PUBLIC HEALTH RELEVANCE: Treatment of myelodysplastic syndromes (MDS), a clonal bone marrow failure disease primarily of the elderly, remains a challenge to the clinician. There are now only 3 FDA approved drugs for this disease. One drug, lenalidomide, is highly effective for a specific subtype of MDS in which a chromosomal abnormality is present. In addition, the drug can have a beneficial effect in about 25% of other MDS patients. The challenge is to identify those patients who have a high probability of response before they receive lenalidomide therapy, which have cause serious side effects. Using gene expression profiling of pre-treatment bone marrow aspirates, we have identified a gene expression profile that identifies responders from non-responders in 26 patients. In this proposal we will screen MDS patients who do not carry the chromosomal abnormality (i.e. have a 1:4 chance of responding to the drug) to see whether the expression profile that we have identified can accurately predict response. In addition we will test the efficacy of a more cost-effect and less complex assay the can be used in the clinic.